JP4400146B2 - Spectral image data processing method and spectral image data processing apparatus - Google Patents

Description

  The present invention relates to a spectral image data processing method and a spectral image data processing apparatus, and in particular, a spectral image data processing method capable of efficiently storing spectral image data while maintaining consistency with the use of general image data, and The present invention relates to a spectral image data processing apparatus.

  Spectral reflectance data of a subject can be photographed by a multiband camera having about 5 to 6 types of sensitivity, or a spectral data camera including all visible light and invisible light. The original spectral reflectance can be estimated by processing such a photographing result by various methods.

  (1) As one of such things, the following patent document 1 shows a technique for holding spectral image data as coefficient data based on principal component analysis from the inventors of the present application. In this method, the spectral reflectance of the subject is subjected to principal component analysis, and the principal component and the amount thereof are retained to retain image data. This proposal has an advantage that an apparent color can be estimated even if the illumination light source changes. In other words, by including basic spectral distribution data and spectral distribution contribution data in the color characteristic data, the spectral characteristics are not lost, and the spectral image data is simply included in the combination of instrument signal values. In comparison, less data can be used. And since it has a spectral characteristic, even if a light source is changed in the reproduction destination, an image can be reproduced with high accuracy.

  (2) Also, at Color Forum Japan 2000, Imai et al. Entitled “Preliminary Study on Image Compression in Multi-spectral Spectroscopy”, which has data for luminance components of all pixels, and each frequency for reduced (decimated) images. A method having each spectral component is disclosed.

(3) Further, Patent Document 2 below proposes a method for providing spectral image data optimized for each part of an image. That is, in order to more efficiently compress multidimensional spectral image data using principal component analysis, multidimensional spectral image data is divided into color regions, and a set of spectral image data for each divided region is obtained. Perform principal component analysis. Since spectral image data is compressed by selecting the number of principal components so that a predetermined color reproduction rate is obtained for each region, the data can be compressed more efficiently by reducing data that has little influence on the color reproduction rate. be able to.
Japanese Patent Laid-Open No. 11-275376 (first page, FIG. 1) Japanese Patent Laying-Open No. 2003-32497 (first page, FIG. 1)

  (1) In order to completely reproduce a subject by the method of Patent Document 1 described above, it is necessary to hold a lot of spectral image data as image data, which is not efficient. In addition, there is a problem that it cannot be used as image data having conventional RGB data (for example, sRGB data defined in IEC6196-2-1).

  (2) In the method proposed in the above Color Forum Japan 2000, redundancy is still left because of the combination of the luminance component and the spectral image data of each wavelength. That is, since the luminance component does not necessarily match the result of the principal component analysis, there is a problem that efficiency is lowered. In addition, in order to make a color image, spectral calculation processing applied to the light source is necessary, and a special reading unit is necessary for a general user who does not care about excessive colors, which causes a problem that the burden is large.

  (3) Although the method of Patent Document 2 described above attempts to optimize by adjusting the type and the number of region division + principal components, there is a problem that it differs from the characteristics that affect the visual perception of an image. That is, there is a problem in which it is not taken into account that the color discrimination ability is lowered where the color change changes finely and greatly.

  (4) Although common to the above (1) to (3), the subject has a self-luminous material and a reflective material, and the reflective material is affected by changes in the light source. Since there is no influence of the light source except for the reflection, there is a problem that an incorrect answer is shown if the light source distribution is uniformly superimposed on all image data.

  It is an object of the present invention to realize a spectral image data processing method and a spectral image data processing apparatus that are consistent with general use and that can efficiently store spectral data including spectral image data. To do.

  More specifically, the object of the present invention is as follows.

  (1) Spectral image data processing method and spectral image data processing apparatus capable of maintaining spectral image data efficiently by devising a method of holding a component having low recognition ability and a component having high recognition ability by human vision It aims at realizing.

  (2) Spectral image data processing that can be handled in the same manner as normal image data when processing is performed in a standard state, and that can use additional information (extended information) as necessary. It is an object to realize a method and a spectral image data processing apparatus.

  (3) Spectral image data processing method capable of obtaining an accurate value for each of the self-luminous object and the reflective object when obtaining the apparent color of the spectral image data under the influence of the light source by giving information on the light source It is another object of the present invention to realize a spectral image data processing apparatus.

  That is, the present invention as means for solving the problems is as described below.

(1) The invention described in claim 1 is a spectral image data processing method for holding spectral data including spectral image data, wherein the input spectral image data corresponds to all images by component analysis. Spectral image data input by the reduction means while generating main basic color data and principal component data of RGB, sRGB, XYZ, or L * a * b * represented by a plurality of main basic colors Generating spectral intensity data corresponding to either a partial image constituting a part of the entire image, a thinned thinned image, or a reduced image generated by reducing the entire image, and the main basic color data, The spectral image data processing method is characterized in that image data is held so as to include principal component data and the spectral intensity data .

According to a fourth aspect of the present invention, there is provided a spectral image data processing apparatus that performs processing for holding image data including spectral image data, wherein the input spectral image data is converted into all images by component analysis. corresponding, RGB expressed in several major basic colors, sRGB, XYZ, or, L * a * b * color reproduction means for generating a primary base color data and the main component data is, spectroscopy is the input Reduction means for generating spectral intensity data corresponding to either a partial image constituting a part of the whole image, a thinned thinned image, or a reduced image generated by reducing the whole image; a main base color data, and the main component data, said synthesizing means synthesizes and either partial image or thinned image or reduced image to generate a composite image data files, to have a A spectral image data processing apparatus according to symptoms.

  In these inventions, when storing image data including spectral image data, main basic color data represented by a plurality of main basic colors corresponding to all images, and a partial image, a thinned image, or a reduced image Image data is held so as to include spectral intensity data corresponding to any one of them.

  For this reason, it is possible to efficiently maintain the spectral characteristics of the subject.

  (2) The invention according to claim 2 is the spectral image data processing method according to claim 1, characterized in that the main basic color data holds a high contribution ratio of spectral principal components. .

  The invention according to claim 8 is characterized in that the color reproduction means generates and holds a high contribution ratio of a spectral principal component for the main basic color data. An image data processing apparatus.

  In these inventions, the main basic color data is held with a high contribution ratio of spectral principal components.

(2) The invention according to claim 2, the image data to be the holding, as spectral intensity data extension information, having identification labels showing another self luminous object and the reflective material, it in claim 1, wherein The spectral image data processing method described .

Further, an invention according to claim 5, as a frequency of light intensity data extension information, further comprising an object type identifying means for generating an identification label showing another self luminous object and reflector, said combining means, said spectral intensity the identification label as a data expansion information further synthesized to generate the synthesized image data file, it is the spectral image data processing apparatus according to claim 4, characterized in.

  In these inventions, when main basic color data is held as a normal image portion with respect to image data, the spectral intensity data expansion information includes an identification label indicating whether the self-luminous material is different from the reflective material.

( 3 ) In the invention according to claim 3 , the spectral image data is generated by photographing a subject with a spectral image capturing means, and the subject is a self-luminous object or a reflecting object at the time of photographing. by determining whether the spectral image according to claim 1 as spectral intensity data extension information, generates identification label showing another self luminous object and the reflective object based on the result of the determination, characterized in that Data processing method.

According to a sixth aspect of the present invention, the spectral image data is generated by photographing a subject with a spectral image capturing unit, and whether the subject is a self-luminous object or a reflective object at the time of photographing. determine, further comprising a determining means for generating an identification label showing another self luminous object and reflector, said combining means, the composite image data further synthesized to the identification label as the spectral intensity data extension information generating a file, that the spectral image data processing apparatus according to claim 4, characterized in.

  In these inventions, when the main basic color data is held as the normal image portion of the image data, it is determined whether the subject is a self-luminous object or a reflective object at the time of shooting, and self-luminous light is used as spectral intensity data expansion information. An identification label indicating whether the object is different from the reflector is provided.

( 4 ) The invention according to claim 7 is the main basic color data corresponding to all images and represented by a plurality of main basic colors, such as RGB, sRGB, XYZ, or L * a * b * ; A composite image data file synthesized so as to include spectral intensity data corresponding to either a partial image constituting a part of the entire image, a thinned thinned image, or a reduced image generated by reducing the entire image. A spectral image data processing method for obtaining an image value under a designated light source, wherein the composite image data file is separated into main basic color data and spectral intensity data, and the spectral intensity data and information on the designated light source The reduced primary basic color data under the specified light source is calculated from the specified light source, and the reduced primary basic color data and the reduced primary basic color under the specified light source are obtained by reducing the primary basic color data to obtain reduced primary basic color data. Determine the specific from the correction coefficient and over data, the image values under the specified light source by performing a correction calculation by the correction coefficients corresponding to the correction factor to the primary base color data to all the image enlarged by interpolation This is a spectral image data processing method characterized in that it is obtained.

Further, the invention described in claim 8 corresponds to all images and is represented by a plurality of main basic colors, such as RGB, sRGB, XYZ, or L * a * b *. From a composite image data file synthesized to include spectral intensity data corresponding to either a partial image constituting a part of the image, a thinned thinned image, or a reduced image generated by reducing the entire image A spectral image data processing apparatus for obtaining an image value under a designated light source, wherein the composite image data file separating means for separating the composite image data file into main basic color data and spectral intensity data; and the spectral intensity data; Designated light source reduced primary basic color data value calculating means for calculating a reduced primary basic color data under a specified light source from the specified light source information, and reducing the primary basic color data by reducing the primary basic color data. A reduction unit for obtaining the primary base color data, and the correction coefficient calculating means for obtaining a correction coefficient from the ratio of the reduced primary base color data and the designated light source under reduced major basic color data, the total expanded by interpolation the correction factor An enlargement unit that generates a correction coefficient corresponding to an image, and a correction operation unit that performs a correction operation on the main basic color data using the correction coefficient generated by the enlargement unit to obtain an image value under a designated light source; And a spectral image data processing device.

In these inventions, an image is included so as to include main basic color data represented by a plurality of main basic colors while corresponding to all images, and spectral intensity data corresponding to a partial image, a thinned image, or a reduced image. If you keep the data,
The composite image data file is separated into main basic color data and spectral intensity data, and reduced primary basic color data under a specified light source under a specified light source is calculated from the spectral intensity data and information of the specified light source, The basic color data is reduced to obtain reduced primary basic color data, a correction coefficient is obtained from the ratio of the reduced primary basic color data and the reduced primary basic color data under the designated light source, and the correction coefficient for the primary basic color data Thus, the correction calculation is executed to obtain the image value under the designated light source.

  As described above, according to the present invention, the following effects can be obtained.

(1) In the invention described in claim 1 and claim 4 , when holding image data including spectral image data, main basic color data represented by a plurality of main basic colors corresponding to all images And the spectral data corresponding to either the partial image, the thinned image, or the reduced image.

  For this reason, it is possible to devise a way of holding the component having low recognition ability and the component having high recognition ability by human vision, and efficiently maintain the spectral image data.

  For this reason, it is possible to efficiently maintain the spectral characteristics of the subject.

( 2 ) In the invention according to claim 2 and claim 5 , when main basic color data is held as a normal image portion for image data, as a spectral intensity data extension information, a distinction between a self-luminous object and a reflecting object is made. An identification label is provided.

  For this reason, by using the identification label, when the apparent color under the influence of the light source is obtained by giving the information of the light source, it is possible to distinguish the part that is affected by the illumination from the part that is not. Therefore, it is possible to obtain an accurate value for each of the self-luminous material and the reflective material.

( 3 ) In the inventions according to claims 3 and 6 , when main basic color data is held as a normal image portion for image data, it is determined whether the subject is a self-luminous object or a reflecting object at the time of photographing. Discrimination is performed, and as the spectral intensity data expansion information, an identification label indicating whether the self-luminous material is different from the reflective material is provided.

  For this reason, by using an identification label based on the result determined at the time of photographing, more accurate information can be used, and information on the light source is given to obtain the apparent color of the spectral image data under the influence of the light source. In this case, it is possible to distinguish between a part that is affected by illumination and a part that is not so, and it is possible to obtain an accurate value for each of the self-luminous object and the reflective object.

( 4 ) In the inventions according to claims 7 and 8 , any one of main basic color data represented by a plurality of main basic colors and corresponding to all images, a partial image, a thinned image, or a reduced image When the image data is held so as to include the spectral intensity data corresponding to the above, the composite image data file is separated into main basic color data and spectral intensity data, and the spectral intensity data and the information on the designated light source are used. Reduced primary basic color data under the designated light source is calculated, reduced primary basic color data is obtained by reducing the primary basic color data, reduced primary basic color data and reduced primary basic color data under the designated light source, Then, a correction coefficient is obtained from the ratio, and a correction operation is performed on the main basic color data using the correction coefficient to obtain an image value under a designated light source.

  For this reason, when processing is executed in the standard state, it can be handled in the same way as normal image data, but by using additional information (extended information) as necessary, the image value under the designated light source can be changed. It becomes possible to ask. Therefore, it is possible to efficiently maintain and use the spectral characteristics of the subject.

  The best mode for carrying out the present invention will be described below in detail with reference to the drawings.

  The best mode spectral image data processing method and spectral image data processing apparatus for carrying out the present invention will be described with reference to the drawings.

  First, referring to FIG. 1, the spectral image data processing method of the best mode for carrying out the present invention and the imaging portion of the entire spectral image data processing apparatus will be described. FIG. 1 is an example of a processing apparatus used in the best mode for carrying out the present invention, and shows a block diagram when image data is held including spectral image data.

  Reference numeral 100 denotes a spectral image capturing unit that generates spectral image data by capturing a subject. Spectral image data obtained by imaging with the spectral image capturing means 100 is sent to the spectral image data processing apparatus 200.

  A spectral image data processing apparatus 200 is a characteristic part of the present embodiment. The spectral image data processing apparatus 200 includes various elements to be described later, and includes a spectral image data processing method for holding image data including spectral image data. It is also a device to realize.

  Reference numeral 210 denotes light source generating means for generating information on a light source such as D65.

  220 is a principal component by performing principal component analysis from principal image data (sRGB image data, etc.) corresponding to all images and represented by a plurality of principal basic colors from the spectral image data from the spectral image capturing means 100. And data.

  Reference numeral 230 denotes reduction means, which is spectral intensity data (reduced spectral image data) corresponding to either a partial image constituting a part of the entire image, a thinned thinned image, or a reduced image generated by reducing the entire image. ) Is generated.

  Reference numeral 240 denotes a self-luminous material / reflecting material discriminating means for generating an identification label indicating the distinction between the self-luminous material and the reflective material as extended information of spectral intensity data.

  250 synthesizes the principal basic color data, the principal component data, reduced spectral intensity data (reduced spectral image data) corresponding to either a partial image, a thinned image, or a reduced image, and, if necessary, an identification label. Thus, a combining means for generating a combined image data file.

  Reference numeral 300 denotes a composite image data file generated by the spectral image data processing apparatus 200, and includes main basic color data, principal component data, and reduced spectral image data corresponding to a partial image, a thinned image, or a reduced image, Further, an identification label is synthesized as necessary.

  As shown in FIG. 2, this composite image data file includes at least main basic color data (RGB, XYZ, L * a * b *, etc.) of a normal image generated by the color reproduction means 220 as shown in FIG. )), Reduced spectral image data (FIG. 2B) corresponding to either a partial image, a thinned image, or a reduced image, and an identification label (FIG. 2C) as necessary. ing.

  The file structure of the composite image data file is as shown in FIG. That is, the normal image portion has image data suitable for general use, for example, sRGB image data. The spectral intensity data expansion portion includes principal component data generated by principal component analysis by the color reproduction unit 220, reduced spectral image data, and subject type identification labels.

  The operation of the embodiment of the present invention will be described below.

(1) Imaging:
First, the subject is imaged by the spectral image capturing means 100. The spectral image data obtained by imaging with the spectral image imaging means 100 is sent to the spectral image data processing device 200. Here, various methods such as multiband photography exist. It is possible to use a normal camera for high-definition images and a mechanism for capturing low-definition images with a spectroscopic camera. In addition, a spectral image can be captured as a high-definition image as it is, and a low-definition image is created by averaging and thinning processing. May be.

  Here, the color reproduction means 220 that has received the spectral image data performs main component analysis by performing principal component analysis on main basic color data (sRGB image data, etc.) corresponding to all images and represented by a plurality of main basic colors. And component data.

(2) Principal component analysis:
Here, FIG. 4 shows the result of principal component analysis using a color chart called Macbeth color checker as an example. Here, holding of principal component data as a result of principal component analysis will be described.

  Most of the information of the spectral image data is kept higher as the lower-order (principal) main component is higher, while the higher-order (non-major) main component is lower in the contribution ratio for accurate spectral reproduction. In general, the low-order main component often approaches a DC component (luminance component).

So, to simply hold the image data,
(2a) Spectral component data: When 400 to 700 nm is held at 10 nm intervals, 31 × spectral component (word),
(2b) Each spectral component intensity data: number of pixels × spectral component number (word),
It becomes.

For example, assuming that the number of pixels is 1536 × 1024, and the number of spectral components is 6,
31 × 6 + 1536 × 1024 × 6 = 9,437,370 (words), which is an enormous amount of data and is not efficient. Here, the word typically refers to about 1 to 4 bytes, and may be an integer type or a floating point type.

(3) Omission of image data:
As a characteristic (visual characteristic) of the human eye, the visual ability is low in the high frequency part, and even if it is slightly different, the difference cannot be distinguished. However, the low frequency portion has a high visual recognition ability and has a characteristic that can clearly distinguish a slight difference. Also, colors that are familiar to the user, such as skin color, have higher discrimination ability than other colors.

  For this reason, information on specific illumination or representative principal components is held for the total number of pixels, and higher-order spectral components are reduced or held for image components of specific portions.

  Alternatively, tristimulus values under specific illumination or values derived therefrom (eg, RGB) are retained for the total number of pixels, and spectral image data is retained for reduced or specific portions of image components. .

As a thinning method, the following method can be used.
(3a) sub-sampling after averaging of surrounding pixels,
(3b) Only the main subject (ROI: Region Of Interest) or the density of this portion is increased.

Then, in order to have all pixels have normal RGB image data and have spectral image data as an image component of a reduced or specific portion, for example, assuming that the number of pixels is 1536 × 1024, the number of spectral components is If it is 6,
31 × 6 + 1536 × 1024 × 3 + 1536/4 × 1024/4 × 6 = 5, 308, 602 (words), and information on spectral image data can be added to RGB image data without a large burden.

  In this case, normal JPEG compression or JPEG2000 is used for the RGB image data, and if the spectral image data part is attached in the form of a private marker or a new tag (in the case of TIFF), this spectral image can be obtained with normal JPEG reading software. The data part is ignored and read.

  Therefore, a general user who does not need to reproduce spectral image data can handle it as normal RGB image data. On the other hand, a user who needs to reproduce spectral image data can calculate and use a color estimated from spectral image data and arbitrary illumination light data as necessary.

  That is, when processing is performed in the standard state, it can be handled in the same manner as normal image data, and additional information (extended information) can be used as necessary. An image data processing apparatus can be realized.

As a modified example, when all pixels have a main component of data and have reduced or higher-order spectral components with respect to an image component of a specific portion, the fourth to sixth parts of the principal component data are reduced to ¼. Then
31 × 6 + 1536 × 1024 × 3 + 1536/4 × 1024/4 × 3 = 5,013,690 (words), further compressed, and spectral image data information can be added to RGB image data without a large burden It becomes like this.

(4) Color estimation under different light sources:
Next, a spectral image data processing method and a reproduction operation (color estimation under different light sources) of the best mode for carrying out the present invention and the spectral image data processing apparatus will be described with reference to FIG. . FIG. 5 is an example of a processing apparatus used in the best mode for carrying out the present invention, and shows a block diagram at the time of reproduction when image data is held including spectral image data.

  Reference numeral 300 denotes a composite image data file generated by the spectral image data processing apparatus 200 (see FIG. 1), which corresponds to main basic color data, principal component data, and a partial image, a thinned image, or a reduced image. Reduced spectral image data and, if necessary, subject type identification labels are synthesized.

  A spectral image data processing apparatus 400 obtains an image value under a designated light source from the above-described composite image data file. The spectral image data processing apparatus 400 is compatible with all images and represents main basic color data represented by a plurality of main basic colors, and all images. From a composite image data file synthesized to include spectral intensity data corresponding to either a partial image constituting a part of the image or a thinned thinned image or a reduced image generated by reducing the entire image, This is a spectral image data processing device for obtaining an image value under a designated light source.

  Reference numeral 410 denotes composite image data file separation means for separating components constituting the composite image data file, corresponding to all images, and main basic color data (sRGB image data) represented by a plurality of main basic colors, and all Spectral intensity data (reduced spectral image data) corresponding to either a partial image constituting a part of the image, a thinned thinned image, or a reduced image generated by reducing the entire image, and a subject type identification label And separated.

  420 is a designated light source generating means for generating data on a designated light source. Upon receiving an identification label of a subject type, an E light source (equal energy light) is applied to the self-luminous object or is used as it is, and is used as a reflection object. Performs processing by giving data corresponding to the designated light source, and obtains RGB values via tristimulus values.

  Reference numeral 430 denotes RGB value calculation means, which calculates RGB image values (reduced main basic color data under a specified light source) using data about a specified light source as separated reduced spectral image data (spectral intensity data). . Reference numeral 440 denotes a reduction unit that reduces the primary basic color data (sRGB image data) to obtain reduced primary basic color data.

  Reference numeral 450 denotes comparison / coefficient calculation means, which calculates a correction coefficient from the ratio between the reduced primary basic color data and the reduced primary basic color data under the designated light source. Reference numeral 460 denotes an enlargement unit that enlarges the correction coefficient from the comparison / coefficient calculation unit 450 and generates correction coefficients corresponding to all pixels. Reference numeral 470 denotes correction calculation means, which performs correction calculation on the main basic color data using the correction coefficient to obtain an image value under a designated light source.

  In the spectral image data processing apparatus 400 configured as described above, main basic color data corresponding to all images and represented by a plurality of main basic colors, and a partial image or a thinned out part of the entire image. When obtaining an image value under a designated light source from a synthesized image data file 300 synthesized to include spectral intensity data corresponding to either the thinned image that has been reduced or the reduced image generated by reducing the entire image The following processing is executed.

  (4a) The composite image data file separation unit 410 corresponds to all images in the composite image data file, and main basic color data (sRGB image data) represented by a plurality of main basic colors and a part of all images Are separated into spectral intensity data (reduced spectral image data) corresponding to either a partial image, a thinned thinned image, or a reduced image generated by reducing the entire image, and a subject type identification label. To do.

  (4b) The reduction means 440 obtains reduced main basic color data by reducing the main basic color data (sRGB image data).

  (4c) The designated light source generating means 420 for generating data about the designated light source receives the identification label of the subject type, gives an E light source (equal energy light) to the self-luminous object, and designated to the reflecting object Given light source data, process.

  (4d) The RGB value calculation means 430 calculates the RGB image value (reduced primary basic color data under the designated light source) using the data for the designated light source as the separated reduced spectral image data (spectral intensity data). To do.

  (4e) The comparison / coefficient calculation means 450 obtains a correction coefficient from the ratio between the reduced primary basic color data and the reduced primary basic color data under the designated light source. This correction coefficient may be obtained as a difference in addition to the ratio.

That is,
CR = Rspect / R'sRGB,
CG = Gspect / G'sRGB
CB = Bspect / B'sRGB,
Calculate

  Here, CR to CB are correction coefficients, Rspect to Bspect are R to B values calculated from the reduced spectral image data, and R'spect to B'sRGB are R to B when reduced from the original sRGB image. Value.

  (4f) The contents of the correction coefficients CR to CB obtained in the above (4d) are enlarged by the enlargement means 460 by interpolation calculation to generate correction coefficients corresponding to all the pixels. As an interpolation calculation of this difference, bilinear interpolation, bicubic interpolation, and nearest neighbor interpolation can be used.

(4g) The correction calculation means 470 applies the correction coefficient obtained in (4f) above to the original sRGB image data, and obtains an image value indicating a color under a different light source. That is, here, the following correction calculation processing is executed.
RsRGB-new = CR ・ RsRGB,
GsRGB-new = CG ・ GsRGB,
BsRGB-new = CB ・ BsRGB
With this correction calculation process, the RGB value of each pixel under an arbitrary light source can be obtained. In this correction calculation processing, the RGB ratio is simply used. However, as in the von Kries model, for example, the cone response (LMS) may be converted once using gradation conversion or matrix conversion.

  It should be noted that the portion labeled with the self-luminous object by the subject type identification label is processed as it is or by applying an E light source (uniform energy light).

  As described above, it is possible to obtain an image value under a designated light source from the composite image data file 300 by using additional information (extended information) as necessary. Therefore, it is possible to efficiently maintain and use the spectral characteristics of the subject. Further, when processing is executed in a standard state without using additional information, it can be handled in the same manner as normal image data.

  In the embodiment described above, the principal component analysis characteristics of spectral image data generally have a first-order term as an intensity component and a second-order term and later as a color component, and the higher the order, the lower its contribution rate. There is a tendency. In addition, the human eye is mainly sensitive to light and dark gradations, while the sensitivity to color changes is reduced.

  Therefore, in the spectral image data processing for holding the image data including the spectral image data, the number of bits held for the higher-order spectral intensity data among the plurality of main basic spectral intensity data is reduced. For example, primary: 8 bits, 2-3: 6 bits, 4-6: 4 bits, etc. are preferable.

  In the above embodiment, when photographing a subject including a self-luminous material, the self-luminous material / reflecting material discriminating means 240 discriminates whether it is a self-luminous material or a reflective material by one of the following methods. It becomes possible.

  (1) A change in luminance is examined during imaging, and a portion where flicker (periodic luminance change) occurs can be estimated as a self-luminous object. That is, for the readout result from the image sensor of the spectral image capturing means 100, the region where the brightness observed in each frame changes is determined as the region of the self-luminous object.

  (2) If there is a steep part in the spectral response after calculating the spectral reflectance and removing the influence of the illumination light, the spectrum is likely to be emitted, so that it can be identified as a self-luminous substance.

  (3) When imaging, the case where the auxiliary light of the imaging means is used is compared with the case where the auxiliary light is not used, the spectral distribution is examined, and if there is no influence of the auxiliary light, it is determined as a self-luminous object, and the influence of the auxiliary light If there is, it is determined as a reflector. In this case, this method cannot be used when the distance to the subject at the time of imaging is long and the auxiliary light does not reach the target sufficiently. Therefore, it is desirable to use this method together with the method (2). Alternatively, when the distance is long, it may be determined as a self-luminous object that is not affected by illumination.

  Based on the determination results (1) to (3) above, the self-luminous object / reflecting object determination unit 240 generates an identification label of the subject type and includes the identification label in the composite image data file 300. This can be realized by setting the value of the low-definition spectral image data to zero or substituting a value far from the high-definition image.

  Alternatively, the degree of approximation (or estimated probability) between the self-luminous object and the reflecting object can also be indicated.

  As a result, when the color estimated from the values of high-definition image data and low-definition spectral image data differs greatly in one image data file, the color of the high-definition image data, that is, the color of the self-luminous material is used. Can be calculated.

  Further, when the subject is a fluorescent color, the spectral intensity in the vicinity of 400 nm to 500 nm is generally higher than that in other wavelength regions. Therefore, this portion may be detected and an identification label may be attached as the fluorescent color.

  In the above embodiment, the present invention can be applied not only to a two-dimensional image but also to three-dimensional object data of an object. In this case, the object shape is held as a vector, and the higher-order spectroscopic portion is adjusted in the thinned portion of the vector.

  In the above embodiment, the conventional image compression method (JPEG, JPEG2000, etc.) is applied to allocate the main component for luminance and the spectral component data of other orders for chromaticity. A compression device can be used.

It is a block diagram which shows the electrical structure of the best form which implements this invention. It is explanatory drawing which shows an example of the content of the image data of the best form which implements this invention. It is explanatory drawing which shows an example of the content of the image data file of the best form which implements this invention. It is explanatory drawing which shows an example of the mode of the principal component analysis of the best form which implements this invention. It is a block diagram which shows the electrical structure of the best form which implements this invention.

Explanation of symbols

100 Spectral Image Imaging Unit 200 Spectral Image Data Processing Device 300 Composite Image Data File 400 Spectral Image Data Processing Device

Claims (8)

A spectral image data processing method for holding image data including spectral image data,
The component analysis of the input spectral image data, RGB expressed in several major basic colors corresponds to the entire image, sRGB, XYZ, or, L * a * b and the main basic color data is *, the main In addition to generating component data,
The spectral intensity data corresponding to one of the input spectral image data reduced image generated by reducing the thinned image or the entire image has been partial image or thinning constitutes a portion of the total image by reduction means Generate
Holding image data to include the primary basic color data, the principal component data and the spectral intensity data ;
The spectral image data processing method characterized by the above-mentioned. In the image data to be held, as spectral intensity data expansion information, it has an identification label that indicates the self-luminous object and the reflective object,
The spectral image data processing method according to claim 1. The spectral image data is generated by photographing a subject with a spectral image capturing unit, and spectral intensity data extension information is determined by determining whether the subject is a self-luminous object or a reflective object at the time of photographing. As an identification label indicating the distinction between the self-luminous object and the reflecting object based on the result of the determination ,
The spectral image data processing method according to claim 1. A spectral image data processing apparatus that performs processing for holding image data including spectral image data,
Main spectral data and principal component data of RGB, sRGB, XYZ, or L * a * b * represented by a plurality of main basic colors corresponding to all images by component analysis of the input spectral image data Color reproduction means for generating
Spectral intensity data corresponding to either the partial image constituting a part of the entire image, the thinned-out thinned image, or the reduced image generated by reducing the entire image is generated from the input spectral image data. Reduction means,
Combining means for combining the main basic color data, the principal component data, and the partial image or the thinned image or the reduced image to generate a combined image data file;
Spectral image data processing apparatus characterized by having a. As a divided light intensity data extension information, further comprising an object type identifying means for generating an identification label showing another self luminous object and reflector,
The synthesizing unit further synthesizes the identification label as the spectral intensity data extension information to generate the synthesized image data file.
The spectral image data processing apparatus according to claim 4 . The spectral image data has been generated by photographing an object by spectral image pickup means, subject to determine a reflective object or a self-luminous object during photography, reflector and self luminous object A discriminating means for generating an identification label indicating the distinction of
The synthesizing unit further synthesizes the identification label as the spectral intensity data extension information to generate the synthesized image data file.
The spectral image data processing apparatus according to claim 4 . Main basic color data corresponding to all images and represented by a plurality of main basic colors, such as RGB, sRGB, XYZ, or L * a * b *, and a partial image constituting a part of the entire image or Spectroscopy that obtains image values under a specified light source from a composite image data file that is composed to include either a thinned image that has been thinned out or spectral intensity data that corresponds to either a reduced image generated by reducing the entire image. An image data processing method,
Separating the composite image data file into main basic color data and spectral intensity data;
From the spectral intensity data and the information on the designated light source, calculate the reduced primary basic color data under the designated light source under the designated light source,
Reducing the primary basic color data to obtain reduced primary basic color data;
A correction coefficient is obtained from the ratio between the reduced primary basic color data and the reduced primary basic color data under the designated light source,
An image value under a designated light source is obtained by executing a correction operation with a correction coefficient corresponding to all the images obtained by enlarging the correction coefficient by interpolation calculation with respect to the main basic color data.
The spectral image data processing method characterized by the above-mentioned. Main basic color data corresponding to all images and represented by a plurality of main basic colors, such as RGB, sRGB, XYZ, or L * a * b *, and a partial image constituting a part of the entire image or Spectroscopy that obtains image values under a specified light source from a composite image data file that is composed to include either a thinned image that has been thinned out or spectral intensity data that corresponds to either a reduced image generated by reducing the entire image. An image data processing device,
A composite image data file separating means for separating the composite image data file into main basic color data and spectral intensity data;
Designated light source reduced primary basic color data value calculation means for calculating a specified light source reduced primary basic color data under a specified light source from the spectral intensity data and the specified light source information;
Reduction means for reducing the primary basic color data to obtain reduced primary basic color data;
Correction coefficient calculation means for obtaining a correction coefficient from the ratio between the reduced primary basic color data and the reduced primary basic color data under the designated light source;
Enlarging means for enlarging the correction coefficient by interpolation to generate a correction coefficient corresponding to all images,
Correction calculation means for performing correction calculation on the main basic color data using the correction coefficient generated by the enlargement means to obtain an image value under a designated light source; and
Min optical image data processing apparatus comprising the.

Images